I run FreeBSD on a Thinkpad X61.
I like the X61 because it has a physically small footprint. Also because I was able to pick up several of them at a good price, so that, with regular backups, my laptop, as well as the data upon it, are mirrored.
However, for several years I have noticed that the X61 overheats and shuts down when I push it too hard. Like, when I am building a new laptop, and running X, and multitasking, and compiling a kernel, and using the ports collection to build things from source, night and day, for three or four days. Somewhere in there it starts warming up and it just accumulates heat faster than it can get rid of the heat, until it shuts down.
I'm kind of disappointed. You would think they would anticipate that in the design.
FreeBSD has some registers in the kernel that can be monitored, much as Linux offers via /proc, to see what the actual temperature is on a per-core basis. I wrote a script and ported it from version to version of FreeBSD to monitor that. I began using powerd(8). That seemed to fix the problem; but at the cost of throttling the CPU to about 75% of its rated capacity.
More recently I upgraded FreeBSD again, and switched window managers, as well. I am still using powerd(8), but the laptop has started overheating again.
I modified a script I use to monitor network connectivity and other performance attributes, to check the temperature, and to use espeak(1) to let me know when the laptop is overheating.
I'm getting woken up at 3 AM: "Dave, my CPU is overheating."
I'm getting into the habit of suspending the laptop when I go to sleep, just to keep from being woken up. Maybe that's not a bad idea; if I'm not actually DOING anything, suspending the machine shrinks my window of vulnerability by 8 hours, every day, and saves power, too.
But it rankles. I should be able to leave a machine up for an entire year without problems. Right?
And a laptop with four or eight cores that are running even FASTER cannot possibly be any cooler than my Thinkpad, today, not without more fans, and shorter battery life.
Windows doesn't seem to have this problem, even when gaming. Or does it? Maybe the only thing that's getting hot is the GPU - and, on gaming laptops, they've addressed that.
Should I give up on using FreeBSD for serious mobile computing and just build me an old-school deskside server? They never overheat and they are 'way easier to clean, you don't have to be a !@#$ Swiss watch repairman to replace the fan and it doesn't take a week to reassemble, either.
Or are there laptops out there that combine beefiness with robust cooling?
It would be nice to get an old MILSPEC laptop.
Suggestions?
I'm surely not the only person facing this quandary.
(Score: 1) by fustakrakich on Friday October 18 2019, @03:36AM (8 children)
I'll bet the CPU heat sinks sprung a leak. Look for green copper oxide, or any metal flaking off the heat pipe near the fins.
La politica e i criminali sono la stessa cosa..
(Score: 3, Insightful) by JoeMerchant on Friday October 18 2019, @09:57AM (7 children)
Mine pack up with dust... they can be cleaned, but I've found that if a laptop runs hot it's a losing game - more effort than it's worth to me to clean the fans every 6 months, just get a cooler running processor next time - those have always run until I can't stand to use them anymore because they are so outdated, usually 6+ years. The hotter processors might not go outdated for another couple of years beyond that, but they're just not worth the trouble, or extra expense, to me.
🌻🌻 [google.com]
(Score: 2, Informative) by anubi on Friday October 18 2019, @10:17PM (6 children)
Swing by an air compressor with it and give it a blow. Right in the ventilation slots where the hot air usually exits.
I do this to mine when it starts running hot.
This usually nets me a cloud of dust, and a machine that will need it done again in a couple of years.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Interesting) by JoeMerchant on Saturday October 19 2019, @02:07AM (4 children)
I have a Sony Vaio that I've done the compressed air thing for, it kept it cool for about 3 months - full disassembly cleaning got more like 6 months - it's a particularly hot running machine, discrete graphics and "high end" processor for its day.
On the other hand, when that Vaio went in for warranty service and I needed something to work on, I bought one of the cool runners to have while it was shipping back and forth... that one ran for 6+ years and never needed cleaning. When they run hot they have to move a lot more air, which translates to a lot more dust.
I also have a ~5 year old i5 NUC here that I cleaned with the air compressor a couple of times, but it's major malfunction became the fan bearing - can't blow that out with compressed air, and hardly worth the repair effort for a NUC that old.
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Saturday October 19 2019, @02:43AM (1 child)
Just so you know, if the fan free spins under the pressure of compressed air, the speed and lateral force put on the fan can be enough to damage the bearings. Next time, try holding the fan still so it does not spin while cleaning, and it won't wear out as easily.
(Score: 2) by JoeMerchant on Saturday October 19 2019, @02:22PM
If you look at the construction of a NUC, you'll see the fan is buried deep... thus the high labor cost of servicing.
🌻🌻 [google.com]
(Score: 1) by anubi on Saturday October 19 2019, @02:59AM (1 child)
Mine is a Compaq WMCQ-56.
WalMart Cheapie.
WalMart had huge stacks of them...$299 ea.
When LED backlights and HDTV was first coming out. One of the first WIN7. Over 10 years ago. It still works as good as it ever did. Wish I had bought several of them.
Its still running my Eagle, LTSpice, MathCad, C++ builder, Delphi, and quite a few more. I disk image with Clonezilla to an external USB hard drive.
And have a simple tripwire in my system and configuration files that let me know pretty fast if I have a miscreant program. Once in a blue moon, I step on a bad web page or open a bad attachment. Very rare, but it has happened. By now, I am so ancient I can't run the new software, not do I seem to run the new malware's either.
But the stuff I do have is perpetual license. I can open stuff I did 20 years ago as if I closed it three minutes ago.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2) by JoeMerchant on Saturday October 19 2019, @02:25PM
I did quite a few Atom Net-Tops back in the 2006-8 timeframe, they were great: tiny, low power, quiet, much faster than the 5-7 year old desktops they were replacing, and they were at the end of the XP product cycle.
Automatic updates of Windows XP rendered them useless - if you wanted a functional Atom NetTop, you had to disable all Windows updates - with that they ran like champs. Easy to become a conspiracy theorist when stuff like that happens.
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Saturday October 19 2019, @06:46PM
you better make sure it has a good water separator on it. i use a metro data vac air dealy. the other thing you could do is pipe in cold air from an A/C. that would lower temps considerably.
(Score: 2) by coolgopher on Friday October 18 2019, @03:38AM (12 children)
Well the obvious question here is "what did the laptop start doing in the middle of the night?" Once you understand what's actually happening, you'll have a chance at fixing/preventing/mitigating the issue.
(Score: 0, Funny) by Anonymous Coward on Friday October 18 2019, @04:44AM (11 children)
Most likely it's Trump, Gabbard, or most likely Yang hacking in to steal his emails for Putin. Try running nethogs and checking for Russian up addresses.
(Score: -1, Offtopic) by Anonymous Coward on Friday October 18 2019, @04:53AM (9 children)
This is offensive. If Yang is stealing emails for anyone, its the CCP.
(Score: -1, Offtopic) by Anonymous Coward on Friday October 18 2019, @05:05AM (8 children)
No this was discussed on Twitter. Yang and Gabbard are pushing Kremlin talking points during the debates: https://soylentnews.org/comments.pl?noupdate=1&sid=34144&page=1&cid=908614#commentwrap [soylentnews.org]
They are Putin puppets just like Trump.
(Score: 2, Offtopic) by takyon on Friday October 18 2019, @05:17AM (7 children)
We know you're memeing though.
Try harder, Russkie!
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: -1, Offtopic) by Anonymous Coward on Friday October 18 2019, @05:22AM (6 children)
How is posting Twitter proof memeing?
(Score: 2, Informative) by takyon on Friday October 18 2019, @05:26AM (5 children)
"Twitter pr00f", like a fine 100-proof vodka. 🇷🇺
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: -1, Offtopic) by Anonymous Coward on Friday October 18 2019, @05:39AM (4 children)
Looks like something a Qtard would come up with. Coincidence that was the 17th post to this thread?
(Score: 2) by takyon on Friday October 18 2019, @05:43AM (3 children)
Coincidence? I think not.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Friday October 18 2019, @07:25PM (2 children)
From the queen bee herself:
https://www.youtube.com/watch?time_continue=3&v=mNLJ3i2oRyg [youtube.com]
(Score: 2) by takyon on Friday October 18 2019, @07:44PM (1 child)
Apparently Hillary has a new book out that she co-authored with Chelsea. No wonder the sore loser crawled out of her hole.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Friday October 18 2019, @07:54PM
Nice Kremlin talking point you've got there. Conclusion: Takyon is Putin's puppet.
(Score: 0) by Anonymous Coward on Friday October 18 2019, @05:44PM
By way of the NSA.
(Score: 5, Informative) by sjames on Friday October 18 2019, @03:39AM (21 children)
It's surprising how much dust can accumulate in blowers and on heat sinks. The last laptop I had developed an overheating problem that I resolved by taking it completely apart and thoroughly removing all dust from it. It made a lot more difference than it seemed like it would, and the thorough cleaning kept the overheating away for a few years. The externally visible intake actually didn't look that bad, but the heat sink by the duct outlet was terrible.
(Score: 5, Informative) by Anonymous Coward on Friday October 18 2019, @04:40AM
That was my immediate thought too. What most people don't realize is how dust problems have runaway tendencies. You see, your heat sink works by exchanging heat with the air based on its surface area. As dust collects, it does five things that all cause an escalation in the problem. First, it acts as insulation on the heat sink, thus reducing its ability to conduct the heat to the air. Second, it reduces the air spaces between the fins, thus increasing back pressure and reducing the amount of air (CFM) moving through the heat sink at the same fan speed (RPM). Third, it increases the surface area available to grab more dust in the future, causing the length of time required to add a unit of dust to decrease. Fourth, by raising the ambient temperature in the case and every heat-transfer surface, the effectiveness of the heat pipe, heat sink, thermal compound, etc. is reduced. Fifth, it can cause the heat pipe to off-gas when too hot due to pressure and reduce its effectiveness that way.
In addition, electronic parts' (like the CPU) sensitivity to heat and ability to generate waste heat increases after each round of overheating, due to a combination of factors, and are thus more likely to overheat in the future even after the heat transfer system is fixed.
As a final note, compiling code is infamous for pushing computers to their limit. It is so hard on them that a few computer part manufacturers donate their resources to compile farms and buildbots for major projects or prioritize early releases to those who do. It is so bad, I'd almost suggest having a dedicated computer for compiling. Get a cheap SBC to compile code on, or use a VPS, or get used to under-clocking to the minimum speed before compiling anything for any length of time. Or get a cheap desktop and do it there where heat is less of an issue. Just leave that poor ThinkPad alone!
(Score: 2) by RS3 on Friday October 18 2019, @04:47AM (11 children)
I agree 100%. I think most laptop cooling is marginal or just inadequate when new. They're made as quiet and small as they can be, so a little dust will push it far into the very inadequate range.
There are no filters on air intakes, so the cooling fins and fan blades become the dust collector. I use compressed air and a blow nozzle. I know I know, never do that! Too bad, I will continue to. Never had a problem by doing it. Some dust can really stick inside of fan blades and make them far less effective at moving air, so taking it apart could be necessary, and might need to get in with a small brush- artist's paintbrush, maybe a toothbrush, or new clean acid flux brush. Of course I still use compressed air.
All of the other answers here are great.
(Score: 3, Informative) by JoeMerchant on Friday October 18 2019, @09:59AM (10 children)
I've found that there is a lot of difference in how hot laptops run depending on the heat generated by the chipset. Cooler (slower, no discrete graphics) chipsets can run much cooler, longer, and more reliably than the bleeding edge stuff.
🌻🌻 [google.com]
(Score: 2) by RS3 on Friday October 18 2019, @02:10PM (9 children)
That's interesting. Any feel for which chipsets are hotter or cooler? Starting with AMD or Intel, core 2, 3, 5, 7, 9, DDR2, 3, 4, ... ?
I usually blame it on poorly behaved software.
That said, this here laptop that isn't my best but was, and I still mostly use it, is about 10 years old, Core2 Duo, DDR2, Intel GM965 chipset with integrated graphics, etc. It came with a 1.8GHz CPU and XP and ran well, fan only ran when I did audio / video rendering, or watched a video for more than 20 seconds or so. Fair enough.
I "upgraded" it with a Core2 2.2GHz, and that alone caused the fan to run more often, and I can't say it's any snappier.
I "upgraded" it with Windows 7, and the fan runs more often, even with my aggressive Windows service pruning.
I "upgraded" it with an SSD. It boots faster, but still mostly feels the same. Oh, but the drive gets corrupted from time to time, so I have to run chkdsk /f, so there's much time wasted there.
Bottom line: I might put it back to original CPU, mechanical drive, and XP seemed snappier (again, heavily pruned), but I probably can't run XP unless I figure out how to get software to stop telling me I need to upgrade and refusing to run. Some kind of plugin / driver that fakes Win7 environment. I've spent no time on that.
Fan runs on and off as I'm just typing here. Vivaldi browser, a list of plugins- maybe some of them are churning. Network activity monitor running (NetSpeed Monitor) that is very well behaved, and does not show any unexpected network traffic... I guess I'll put original CPU back and see how that does...
(Score: 1) by nitehawk214 on Friday October 18 2019, @03:12PM (1 child)
Not just the chipset, but the motherboard manufacturer. Two different boards running the same chipset will have different thermal qualities. Depending on how the motherboard is arranged, like not getting the heatsink for the chipset wedged under a video card or too close to other heat-generating components.
"Don't you ever miss the days when you used to be nostalgic?" -Loiosh
(Score: 2) by RS3 on Saturday October 19 2019, @03:47PM
Good points. I've had maybe a dozen laptops open / apart, and some do have a separate heatsink and heat pipe just for the main "chipset" chip, (we need a better term for a single chip that integrates what used to be dozens of chips...) and sometimes the GPU has its own heatsink / fan.
(Score: 3, Informative) by JoeMerchant on Friday October 18 2019, @03:57PM (6 children)
I go by the TDP numbers, if they're not rated to handle more than 15W, generally they won't produce more than 15W...
In the distant past, Intel had specific lower power lines of the Core series for use in Laptops, but not all laptop vendors used them because the specs on the higher power chips are obviously "better." You'll notice that even most NUC applications only have two cores / 4 hyperthreads even in the i7 parts, because they know that tiny little case and fan can't handle 4 cores, and the low profile NUC cases also didn't come with i7s installed, only i3 and i5.
As a very broad generalization, I've found the i5 parts to be a very good compromise of heat performance and compute performance, i7 tends to push harder and hotter. Again, in the 15 year ago timeframe, AMD was winning the low power performance game, but only until Intel noticed them making gains in the server market - one development cycle later Intel was on-par with AMD. I think AMD is making gains again in the flops/watt metric, but when you get down below 20W it's as much about the chipset as it is the processor.
🌻🌻 [google.com]
(Score: 2) by takyon on Friday October 18 2019, @05:35PM (4 children)
You could put 32 cores in a NUC. You'd just have to clock them lower. You can also have different turbo clocks based on the number of maxed cores. They actually list that for the Kaby Lake-R NUC chips [wikipedia.org].
Dual-core NUC CPUs have more to do with Intel being greedy. They were expensive and released during an era when AMD was not providing effective competition on IPC. Looking back, many of these were $300-$400 dual-cores. For many years, Intel released the same ol' dual-cores and quad-cores, and occasionally some pricey 6-8 core enthusiast chips.
Intel's First 8th Generation Processors Are Just Updated 7th Generation Chips [soylentnews.org]
Intel Releases 8th-Generation "Coffee Lake" CPUs, Including Quad-Core i3 Chips [soylentnews.org]
Intel Slashes Price of New 18-Core CPUs in Half [soylentnews.org]
Those days are over. And now AMD is jumping into the NUC trend (first two links are third party attempts):
ASRock Announces iBox-R1000 AMD-Powered 'NUC' [tomshardware.com]
The ASRock DeskMini A300 Review: An Affordable DIY AMD Ryzen mini-PC [anandtech.com]
The desktop is dead, and Intel's NUC killed it: AMD to build a SFF PC kit [techrepublic.com]
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by JoeMerchant on Saturday October 19 2019, @05:46PM (3 children)
I've seen the "AMD NUC" announcements some time ago, but not actual hardware to buy yet.
Of course you could clock your 32 cores slower and cooler, but in practice Intel mostly doesn't ship the 4+ core chips with those low TDP configurations.
Intel did make a low power 8 core (not hyperthreads) chip that I dearly wanted to see upgraded to the next fab shrink, but they never did it, and it was just too outdated in the form they did ship it in to compete.
🌻🌻 [google.com]
(Score: 2) by takyon on Saturday October 19 2019, @06:16PM (1 child)
A interesting option could be the "5-core" Lakefield chips [soylentnews.org] with 5-7 Watt TDP [wikichip.org]. It could be a performance regression depending on what you have, but could also allow a fanless system with decent CPU/GPU performance. And it would be the very first "big.LITTLE" style x86 processor (that I know of). The supposed 18% IPC increase on the single Sunny Cove [wikichip.org] core might help compensate for the lower clock speed.
As I said in the older article, it could be neat to see a push for big.LITTLE (large/small) at Intel and AMD. 5nm and 3nm die shrinks will reduce the die area necessary to throw in extremely tiny integrated graphics or small cores on every SoC-style chip. AMD could pair large Ryzen cores with small Jaguar-style cores. Ratio of big:small to be determined. Intel is doing 1:4, but 1:1 or 8:2 might also make sense.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by JoeMerchant on Sunday October 20 2019, @02:41AM
Currently have some ST Micro H7 "Discovery" boards with this philosophy for low power embedded use - dual processors, one faster and more capable, the other focused on lower power.
🌻🌻 [google.com]
(Score: 2) by takyon on Tuesday October 22 2019, @12:45PM
Hi there. This leak just dropped:
Intel Ghost Canyon NUC With Next-Gen Element Modular PC Board Full Review Leaked [wccftech.com]
Not sure if this is low enough power to be called a true NUC, but there you have it. 45W is also a pretty low TDP for 8 cores, 16 threads, although we know that Intel plays dumb games with TDPs. Compare to 65W TDP for 8-core Ryzen 7 3700X, or 12-core Ryzen 9 3900 (a new OEM-only part).
Looks like Intel Element is more of a real product than believed [soylentnews.org].
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by RS3 on Saturday October 19 2019, @03:43PM
I was very recently given an i5 laptop (by someone I work for), so thanks for the good info re: i5 power consumption. It'll take me a while to start using it- I had to physically clean it a lot, and scans showed dozens of malware. I need to get the guys files off of it, including Outlook dumps, then I'll probably put his HD in an external USB case and give that to him. I remember the P3 being that way- many had no direct fan, and never got hot. I still have some running in servers and they're perfect for it.
(Score: 2) by driverless on Friday October 18 2019, @05:20AM (6 children)
You're using your laptop as if it was a server-grade system. The simplest solution would be that if you're going to be rebuilding your kernel and parts of the ports collection 24/7 for days on end, use a system suited for it, not a laptop.
It's like someone asking why their Porter Cable power tools keep burning out when they're using them in their industrial construction business...
(Score: 2) by driverless on Friday October 18 2019, @05:22AM (3 children)
And the snarky version of that: No, my laptop (Lenovo X1) never overheats, because I use it as a laptop, not a rackmount server.
(Score: 0) by Anonymous Coward on Saturday October 19 2019, @06:48PM (2 children)
and you bought the skinniest, most tightly packed laptop you could find and then you try to run the cpu at 100%. do you even bake bro?
(Score: 3, Insightful) by toddestan on Saturday October 19 2019, @09:48PM (1 child)
The purpose of the laptop is to remote into the powerful server where the real work gets done. And if you're only going to use it as a portable machine, might as well get a small and light one.
(Score: 2) by driverless on Sunday October 20 2019, @12:56AM
That's exactly it. I got my X1 because I don't want to haul a "portable workstation" (snort!) through airports and train stations all over the world. If I need anything done that requires serious processing, I do it on a server (or servers) in a data centre somewhere.
(Score: 2) by takyon on Friday October 18 2019, @05:42AM
It probably can be used that way with a little tinkering. After making sure that power management is being handled correctly, perhaps consider underclocking for a 24/7 system. It's just that laptops made 12 years ago were kinda shit.
Today's cheap $100-200 laptops can outperform various systems made 8, 10, 12, 15 years ago. A fanless system that makes zero noise can outperform the Thinkpad X61, use hardware decoding for modern codecs like VP9/H.265, use more than 4 GB of memory and 64-bit code, etc. I'm seeing 128 GB SSDs in $190 systems, compared to a 100GB (7200rpm) or 160GB (5400rpm) hard disk drive for the ThinkPad X61 [lenovo.com]. The ThinkPad X61 does have an "imbedded numeric keypad"; I'm not sure you can find that on 11.6/12-inch systems anymore.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by sjames on Sunday October 20 2019, @12:45AM
Actually, I was using my laptop as a portable lesser powered workstation. Once I cleaned the dust out, it was fine. That's probably because I opted for a thicker and heavier laptop that was suited to the use I intended to put it to.
(Score: 2) by Bot on Friday October 18 2019, @11:54AM
Same here with an acer aspire.
Using the (inter integrated) graphics would lead to overheat. Took it apart, all seemed well, took apart the fan, lots of dust inside. cleaned it up, end of the probs. They should have small air filters, washable, like some AC units do, but OFC this is planned obsolescence.
Account abandoned.
(Score: 1, Informative) by Anonymous Coward on Friday October 18 2019, @03:47AM (4 children)
Have you taken it apart and applied Artic Silver 5? A lot of the OEM stuff was pretty crappy and that's about a 14 year old laptop.
Also, laptops are NOT meant to be run 100% for any real length of time. Maybe you should be doing your compiling and cpu intensive stuff on a machine designed to be run 100% all the time.
(Score: 2) by JoeMerchant on Friday October 18 2019, @10:16AM
Agree - the MacBook Pro I had from 2006 had defective thermal paste from the factory, fried its GPU. I applied better paste (or maybe just applied the paste better) and it improved dramatically, but not in time to save the GPU from heat death a few months out of warranty.
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Friday October 18 2019, @11:19AM (1 child)
We run MacBook Pros as open directory servers 24/7, and have been for four years.
(Score: 2) by Bot on Friday October 18 2019, @11:56AM
Well maybe the cpu and gpu loads aren't all that high with your setup. See what top, or equivalent, say.
Account abandoned.
(Score: 2) by toddestan on Saturday October 19 2019, @10:03PM
It may be worth checking the heatsink compound and that the heatsink is seated correctly.
I have a Thinkpad R60, so of the same era as your laptop. I still use it - my demands for a laptop are pretty low and I haven't found a replacement I like. A long time ago, I dropped it. Put a decent mark on the case but it otherwise seemed okay. But after that it would overheat when I pushed it hard. I took it apart, and found when I had dropped it I had slightly dislodged the heatsink. Took the heatsink off, replaced the compound with some silver-based compound (may have been Arctic Silver or some knockoff), cleaned it all out while I was in there, and seated the heatsink back to where it was supposed to be. The overheating problems went away.
While the X series was always the smallest and most portable Thinkpad, I would still expect a Thinkpad of that era to be able to run at 100% without overheating assuming everything is still working as it should.
(Score: 4, Informative) by takyon on Friday October 18 2019, @04:03AM (8 children)
Not necessarily true at all. Newer designs on smaller process nodes can have more cores, higher clocks, and less power consumption.
I'm assuming your 2007 Thinkpad X61 has an Intel Core 2 Duo 2.0GHz T7300. 35 Watt TDP, dual core, no boost clock. There are other options but that seems typical.
I have an A6-3400M laptop from 2011. It cost about $450 at the time. It's a 35 Watt TDP CPU, quad core, 1.4 GHz - 2.3 GHz.
I look on Slickdeals for AMD laptops, and find a bunch around $200 with AMD A9-9425, AMD A9-9420e, and AMD A9-9420. These are cheap Stoney Ridge (Excavator) parts from 2016. If you want to splurge, there's the Ryzen 3 3200U in laptops around $270-$300. You can find Intel's i3-8145U in laptops around $230-$350 (looks like the more expensive ones are 2-in-1 touchscreens).
How do these CPUs compare (using PassMark benchmark and clock/TDP stats)? Sorry for no table.
CPU / Year / Core + Clocks / PassMark Single-threaded (Multi-threaded) / TDP
Intel Core2 Duo T7300 / 2007 / 2 cores @ 2 GHz / 711 (1193) / 35W
Intel Core2 Duo T8100 / 2008 / 2 cores @ 2.1 GHz / 829 (1298) / 35W
AMD A6-3400M / 2011 / 4 cores @ 1.4 - 2.3 GHz / 539 (1888) / 35W
Intel Celeron N2830 / 2014 / 2 cores @ 2.16 - 2.41 GHz / 517 (966) / 7.5W (4.5W "scenario design power")
Intel Celeron N3350 / 2016 / 2 cores @ 1.1 - 2.4 GHz / 763 (1111) / 6W (4W "scenario design power")
AMD A9-9425 / 2016 / 2 cores @ 3.1 - 3.7 GHz / 1497 (2501) / 15W (10-25W config.)
AMD Ryzen 3 3200U / 2019 / 2 cores @ 2.6 - 3.5 GHz / 1687 (4775) / 15W (12-25W config.)
Intel Core i3-8145U / 2018 / 2 cores @ 2.1 - 3.9 GHz / 2178 (5550) / 15W (10-25W config.)
"Config." means the laptop manufacturer can configure it to run hotter or cooler. Broadly speaking, lower TDP will correlate to less heat and fan noise. Laptop CPUs under 10W may not even require a fan. To that end, I included a lower TDP Celeron I have, and a newer version.
The Intel Celeron N3350 uses a fraction of the power of the Intel Core 2 Duo T7300, allowing it to be used without a fan, but delivers around the same performance. Newer laptop CPUs have about 3-4 times the performance while at lower TDPs than 35W. This is despite being only dual-core (some have 4 threads). Newer laptop CPUs also tend to have better idle power consumption and other power related features. Also note that the Intel Core2 Duos listed don't have boost/turbo clock functionality. That alone may be a big culprit in causing them to run hot. Although if you are running them 24/7, maybe that doesn't matter.
While there has been a cratering of Moore's law, there have still been significant improvements in performance and energy efficiency in recent years. This is set to continue.
If the cheap crap out there is not interesting to you because of size or build quality (which has arguably improved over the years), you may be interested in this:
It's Alive!: 2007-Era ThinkPad X61 Reborn with Modern Parts [laptopmag.com]
I assume that's a lot more than you paid for your discount ThinkPads.
If FreeBSD is just shitty at handling the fan, well then that's a tragedy.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by takyon on Friday October 18 2019, @10:11AM (1 child)
Here's the faster of the two AMD chips launched in January [soylentnews.org] for fanless and Chromebooks. I was disappointed when I heard about these since I want to see them put 7nm or 14nm Ryzen in this space, not 28nm Excavator:
AMD A6-9220C / 2019 / 2 cores @ 1.8 - 2.7 GHz / 1173 (2071) / 6W
Nevertheless, it's 117% faster than my 2011 machine in single-threaded and 9% faster in multi-threaded, despite having half the cores (both have 1 thread per core) and about 1/6th the TDP. No fan noise. That's progress.
The equivalent Intel chips, like Intel Celeron N4100 [cpubenchmark.net] or Intel Celeron 3965Y [cpubenchmark.net], are a little better or worse.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Friday October 18 2019, @10:57AM
Thanks for all of that information, it's really handy.
Somewhere in my house is an old HP laptop with a Core 2 Duo and 4GB of RAM, and my wife has a Chromebook with a Celeron N3330 (or something like that) and 4GB of RAM, and the general impression is that performance is pretty similar. That seems to line up with what you wrote.
(Score: 3, Interesting) by Hyperturtle on Friday October 18 2019, @02:06PM (4 children)
I can comment, I had similar issues.
I have a laptop, that I still use (not writing on it now, though), that I upgraded from a Core 2 Duo T7200 to a T7600. This generated a lot of additional heat; I ended up causing the heatpipe to leak and render the video, chipset, and CPU cooling entirely ineffective.
I had to buy... a replacement video card (it was an "MXM" style, so I just had to buy the same card), and a new heatpipe. I added copper shims to help transfer heat, I put in higher quality thermal pads, and replaced the thermal material with MX-4.
Then... the second time this happened, I had to replace the system board, because there was a burn mark around the chipset. No amount of MX-4 was going to fix that. Also, the heatsink/heat pipe failed again.
The repairs were cheaper than getting a new laptop (the laptop in question is an older HP NW9440 'desktop replacement' back from 2007 or so--quite modular, quite durable) and I had replaced the 1680x1050 screen with a 1920x1200 screen at the same time I did the system board repair. Mind you, I have had ABSOLUTELY NO EXPERIENCE taking apart laptops (other people in IT always were doing that back when things weren't glued together...) but it worked out well.
Yet... why are these temps going crazy?
It turns out, that the laptop... I had installed windows 7 on it long ago. There was no driver from HP for Windows Vista or Windows 7 that allowed fan control based on reported temperatures. There wasn't even a fan control. It was always spinning at 30% (more on that later). I thought it had been quieter after the OS upgrade from XP, but I had so many other transition problems to worry about that I forgot to look into that.
I ended up buying a $20 "Fan dock" for the laptop; I have what was the 2nd best docking station for it (not a port replicator, but a dock), so I had to choose between one or the other. In the docking station, it was a little cooler because of how it was situated/better airflow, but still. It was way too hot.
The fan dock (5 fans, two speed controls... ) worked *marvelously*, but it still wasn't quite the same as actually specifically cooling the CPU -- which, due to the shared heatpipe contact between the components and radiator, would affect everything as it heated up. The fan dock cooled the entire 'chassis' of the laptop, but nothing specifically got aimed at the real problem. It just helped the chassis shed heat built up internally that couldn't get vented out the right way.
I later found on github a laptop fan control program that was compatible with the HP9440. This made the laptop fan dock entirely unnecessary! YEARS of having this problem, all those broken parts caused by overheating... now it runs just like it used to, with the better CPU...
In addition, to make it all even better... I ended up using "coolaboratory liquid ultra" galliumwhatever TIM/thermal interface material. I removed the MX-4 and added the liquid metal to the CPU and chipset, and then applied some to those copper shims I mentioned and used those with the GPU "MX" card to shed heat further (I had some concerns about the gallium spreading too much, so I went about it conservatively).
The laptop fan control program showed the fan was always at 30% on default; with a profile included, it had a control response curve based on temperatures. The only problem was that the program reports temperatures that are 10C cooler than what it actually is (as based on a temperature "gun" I have for my various computer and other projects...) I had to create a custom fan profile to suit my needs based on an adjustment of 10C compared to what was reported.
Software like speedfan and hardware info monitor etc all reported the temperatures, but never could touch the fan. There just was no way to control it until I found that software on github.
With the changes I made, the fan software, and the docking station, I can run something like boinc on it 24x7 and it doesn't overheat. But... I am not tempting fate like that anymore, but the fact is I never could have used the CPU at even 50% for a prolonged period of time before. Now it all works even better than when it was new.
Whether that amount of effort is worth it for a 13 year old laptop (even with the highest end cpu for the era) is a good question, but it's one of those systems I've dropped and taken everywhere and has lasted and lasted and only had this fan issue. I'm reluctant to part with it and it's got a lot of my "crap something important broke and what do I do" types of tools on or accessible to this -- and the 17" 1920x1200 screen is a real treat now that it all works well enough to use regularly.
I suppose its a good target for replacement with a netbook or something, but.. if I did that I'd have denied myself this frustrating opportunity to be cheap yet so much richer in experience...
Building on this, i have a recent linux build on Debian 10.1 that for the life of me, I cannot find something that can provide administrative control of the CPU fans. They report 0rpm, yet respond to system conditions and get faster slower as needed. I say this because... if I had problems in windows 7 with cpu fan control, and problems with a desktop in linux with cpu fan control (I can even control the raid cage fans but not the cpu fans), I imagine that its possible your laptop is also not quite behaving properly--and my original suggestion of a $20 or $25 laptop "fan dock" might be a good idea to help deal with the issue in the short term while you research how to fix it for the long term.
(Score: 2) by takyon on Friday October 18 2019, @02:45PM (3 children)
I agree with using a system as long as possible. The longer you prolong your next purchase, the better off you should be.
In the near term there should be systems with AV1 hardware decode support (AV2 when?). Hopefully, 1366x768 panel resolution goes away. Typical core counts will increase as AMD raises its top laptop CPU from 4 cores to possibly 8, 10, or 12 cores. Intel will double its integrated graphics performance with "Xe" in 2020-2021. DDR5 support should appear on laptops within 2-3 years. We may see much more powerful ARM laptops, and possibly some niche use of RISC-V. In under a decade, we may see 3DSoCs capable of dramatically increasing performance and efficiency in every form factor. So there are things to look forward to, and there will almost certainly be significant improvements even in 15-20 years, despite the death of more slaw. Although by that time, all laptops may become tablets/2-in-1s.
My 2011 laptop has a messed up screen, dilapidated hinge with a hole in the plastic body, some keys near that hole that need to be hit hard to work, and more fan noise than I'd like. But it still works reasonably well where it sits, and the hard drive is still fine. I'm sure that it is not worth it to try and repair these problems, although I may hit it with the canned air.
At some point, you do have to give up, and there will be impressive improvements available. 4x performance while fanless will happen. If 3DSoC happens, somewhere between 10x-100x is possible.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by acid andy on Friday October 18 2019, @08:05PM (2 children)
THIS. I don't know how people can put up with that in 2019. Well, I do--they just don't know what a resolution is and don't want to know.
If a cat has kittens, does a rat have rittens, a bat bittens and a mat mittens?
(Score: 2) by barbara hudson on Saturday October 19 2019, @02:15AM (1 child)
As for cooling, stick it in the fridge, Ethernet and power through a notch in the door gasket, and ash in as needed. Or outside in winter.
Or use top to find the culprits, and renice them to lower their priority. Just remember to check that another process doesn't become the new cpu hog.
SoylentNews is social media. Says so right in the slogan. Soylentnews is people, not tech.
(Score: 2) by lentilla on Saturday October 19 2019, @05:36AM
nice doesn't work that way.
If you have a process that uses 100% of the CPU and you re-nice it, the process will still take up 100% of the CPU. The only difference is that the nice'd process will yield more readily to other processes with a lower nice value.
Or to put it another way: if you nice a background-running CPU hog, your word processor will feel "snappier", but your CPU will be as hot as it was without all the "nicety".
(Score: 2) by kazzie on Friday October 18 2019, @04:38PM
I've got experience of using a Thinkpad X61s as a daily driver until a year ago, when I switched to an X250. (I've kept the X61s as a Windows XP machine, and use it to talk to retro devices with its serial port through an Ultrabase.)
The X60/60 range do run hot, so you're starting from a bad place anyway. I've found that the Intel 4965 wireless card just under the right palmrest pumps out a lot of heat; thankfully as a left-handed user I don't need to rest my palm there very often. Some people have removed or replaced the wifi card [thinkpads.com] with a later model to cut down on the heat generated. Using a different card requires a BIOS replacement to get rid of the built-in card whitelist.
I feel that the X250 is a better laptop in virtually every regard than the X61s (other than the loss of legacy ports and a slight drop in biuld quality: no longer the legendary IBM standard). It's faster, cooler, and lasts longer on its battery too.
My power settings for the X61s weren't very good under (Slackware) Linux: I never found I got the same amount of battery life that the Windows drivers gave (though I never went further than tweaking things with powertop). The X250 is a lot better, and gives me something approximating a full day's usage on battery power. That's still the case now it's running Arch. Admittedly, I don't thrash it much more than the occasional compilation of packages from source code, but if it's lasting longer on a full battery charge, it must be wasting less of that energy in heat output.
(Score: 5, Informative) by Mojibake Tengu on Friday October 18 2019, @04:47AM (1 child)
should be already present in default kernel, but must be plugged properly
For Intel CPU: coretemp
For AMD CPU: amdtemp
or
check it:
if it works, put into /boot/loader.conf
or
Respect Authorities. Know your social status. Woke responsibly.
(Score: 0) by Anonymous Coward on Friday October 18 2019, @05:15PM
Arigato!
I did not know about coretemp(4).
I should use find(1) more. It complements `man -k`.
However, it only seems to install two additional registers which are actually less precise than those that are already included:
% while 1
while? date && sysctl -a | grep -i temperature
while? sleep 30
while? end
Fri Oct 18 10:09:30 PDT 2019
hw.acpi.thermal.tz1.temperature: 58.1C (the old registers I was already querying)
hw.acpi.thermal.tz0.temperature: 59.1C
dev.cpu.1.temperature: 59.0C (the new registers added by coretemp)
dev.cpu.0.temperature: 59.0C
Many excellent comments and suggestions, thank you all!
(Score: 2) by Arik on Friday October 18 2019, @05:09AM
Next, look at where you have it sitting. Put something underneath, something that doesn't restrict air flow. Just getting the laptop up off the desk surface on a little frame can improve airflow immensely.
Next, open the thing up, it's probably full of dust bunnies and gunk. Clean it out carefully. Remove the CPU if possible, clean it off, apply fresh thermal compound and re-assemble.
If it's STILL overheating after all that, you might have to replace the fan, or supplement it with an external fan.
If laughter is the best medicine, who are the best doctors?
(Score: 2) by shortscreen on Friday October 18 2019, @05:36AM
My laptop (which is 16 years old BTW) does not overheat, because it has a 5.5W CPU. But it has Intel speedstep, which means I can change the voltage and frequency on the fly, aka throttling, just like newer CPUs. Sadly, a lot of systems ship with inadequate cooling, so if you don't want to physically alter it to upgrade the cooling (or point a desk fan at it), then locking the CPU to a lower speed/voltage is one way of avoiding trouble.
BTW, is the cooling fan already running at max speed? Maybe this needs a software fix also.
I don't know what utility you can use on a Core 2 to change CPU speeds, but you can make your own.
set bit 16 in MSR 0x1A0 to enable the feature
read the status from MSR 0x198
write changes to MSR 0x199
the two low bytes contain the frequency ID (multiplier) and voltage ID
(the code has to execute on the primary core to work)
Some Core 2 CPUs have a very limited range of settings. The lowest multiplier is 6, with an option to run the FSB at half if the chipset supports it, and most voltage IDs are not valid (won't be set).
You might be able to replace your CPU with a more efficient one, which either has a lower TDP or has the same TDP at a higher speed (which implies a lower TDP at the same speed). Of course, if it is soldered down rather than socketed it poses a greater difficulty. Sometimes CPU swaps cause trouble with the BIOS.
(Score: 0) by Anonymous Coward on Friday October 18 2019, @05:40AM (1 child)
I have the same problem but under LMDE instead of FreeBSD. I use a taskbar plug-in* which shows the CPU speed and reduce it down to 1.2 or 800 GHz when I'm going to leave it running for days on a CPU intensive task.
*Sorry, am on Windows and can't look up the program
(Score: 2) by takyon on Friday October 18 2019, @05:49AM
Here, I'll correct that for you:
*1.2 GHz or 800 MHz
(Jumping the snark.)
Maybe AC can find a way to underclock automatically during typical sleeping hours, and have it run noisy the rest of the day.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Friday October 18 2019, @06:24AM
I bet the thermal paste has turned into dust by now. Open it up and see what's going on, should be dead simple on an X61. Manufacturers always tend to use the shittiest thermal paste which dries out just after the warranty expires. Get yourself a 20 gram tube of Arctic MX4 and it will last you for years, quite literally.
(Score: 3, Interesting) by lentilla on Friday October 18 2019, @06:33AM
Back in the day I built a stand for my 2006-era Core Duo laptop. Roughly the size of a fat phonebook, it both cooled the laptop and lifted the screen to a more appropriate height for desktop use.
The stand was constructed as a four-sized frame and the frame sat on the table. The laptop sat snugly on the top so that the completed box was mostly air-tight. In the rear of the frame I mounted an large under-volted fan out of a desktop computer with a HEPA filter from a vacuum cleaner. The design relied on the fact that this particular laptop had air inlets below the RAM sockets on the underside of the laptop, and hot exhaust air blew out the sides.
The solution worked quite well. The internal laptop fan rarely kicked in, dust ingress was greatly eliminated and I didn't have to put up with that awful whirring noise from a laptop-sized fan. On the other hand it was painfully obvious that what I really wanted was a desktop computer.
As others have suggested you could programatically throttle your CPU. From your description you mention using powerd with varying levels of success, so I figure you have already scoped out that particular avenue. Being more familiar with Linux, I would have written my own daemon that monitors various temperatures and changed the CPU scheduler accordingly. If you wanted to get fancy you could code a Proportional-Integral-Derivative controller [wikipedia.org] to adjust the target CPU frequency according to system temperature! (That being said, I'm not familiar with how FreeBSD approaches CPU frequency scaling.)
Of course, the correct approach is to obtain a more robust workstation. Laptops; despite their starry-eyed promises; have never seemed completely robust. I suppose there has to be some trade-off for portability.
Failing the application of ever-more-extreme cooling solutions (either hardware- or software- based) my bare-minimum suggestion is to ensure the laptop is on a flat surface and preferentially lift it off the deck by an inch.
Good luck.
(Score: 3, Informative) by ledow on Friday October 18 2019, @07:27AM (2 children)
Any computer that overheats under any use is unreliable. It's that simple. It should be able to thermal manage either its own fans or its own CPU and dial down the speed accordingly, or it's just breaking itself and going to corrupt your filesystems etc. with its hard shutdowns.
I wouldn't rely on it. Not at all. I would leave it overnight, especially. A fire is unlikely but if it's getting hot enough to shut itself down, and has it's own battery, you're risking a fire.
I have a several-years-old gaming laptop. Because of the GPU in it, it is capable of overheating but I really, really have to push it. It's shut down maybe once or twice in its life - when the GPU has been maxed out, the CPU been working hard, it hasn't had a clean in a while, it's sitting on a soft surface, a vent is blocked by something, etc.
It can easily get hot enough to hurt, just from ordinary use and you CANNOT call it a laptop, for good reason. You will end up with reddened thighs from the heat even through thick jeans.
But in everyday use (and it gets heavily used every day and has for every day of its life), it's absolutely fine. The reason you know it's overheating? The games show a symptom - they will go full speed, 60fps... and then they'll drop to 20fps for about ten seconds. The fans are already at max and stay at max, but the CPU and GPU will throttle down. In the Windows event log, you'll see it has messages dialling back several of the cores to a lower clock speed because the firmware is telling it to do so. There's no chance of a serious overheat, just a serious slowdown.
Now, when I'm gaming, and I have the laptop in front of me, and it has my full attention, my workaround is a USB fan I bought on Amazon that's designed to fit on the side-vent of a laptop and pull air out of the laptop vent at 5000rpm. That, on its own, is enough to tip the balance and the CPU/GPU stay at a constant 60fps, with the exhaust air in the 30's (but that's not the hottest part of the laptop, by far - the GPU can hit 80/90 easily). It's noisy, it's unofficial, but it works and it's safe because the computer is safe itself if that's still not adequate or if something goes wrong with the fan connection while I'm gaming.
In summer, that fan is a necessity to play any 3D game which has any kind of serious demand. In winter, it's nice to prevent the occasional slowdown if I'm gaming really hard.
But if your laptop isn't capable of managing the heat on its own, you really shouldn't be using it. Clean it out. Re-silver the heatsink. Replace the fans. Yes. But not just leave it running overnight if it just bails out and gives up and shutsdown. That's gonna screw your components and batteries with the heat, and be a risk.
Buy a new laptop. Chances are it'll wipe the floor with your one without any thermal management issues, and be safer. But any kind of bolt-on solution should not be used to compensate for a laptop that can't manage it's own temperature in any other way than to hit the thermal limit, give up the ghost and shutdown hard.
(Score: 2) by barbara hudson on Saturday October 19 2019, @02:28AM
Nobody uses laptops on buses any more - phones killed that. People put them on tables and desks. Raising the rear with a book as a shin improves airflow and raises the screen.
An external bug screen, mouse, and keyboard makes it a great desktop replacement. If it's heating up and you have pets, clean the fur out. Otherwise clean the dust out.
You can overheat desktops. It's not like desktop boxes are immune from heat.
SoylentNews is social media. Says so right in the slogan. Soylentnews is people, not tech.
(Score: 2) by Common Joe on Saturday October 19 2019, @03:52PM
I came here to say something similar. This is the first comment I saw that said the computers should be self regulating -- throttle down when it gets too hot and process slower until it cools again and then it can throttle up a little bit. It should reach an approximate equalibrium. That is the primary rule. Blowing out the fans from dust won't help the problem where it shuts down due to heat -- it just puts off the inevitable. The shut down mechanism is a last ditch effort by the computer to prevent fires. This means the laptop is not reliable anymore.
I'm going to add to this information.
I'm too lazy to look it up, but a kid found out the i9 in the Apple Laptops (when it first came out, don't know about now) would purposely throttle. He stuck it in the freezer and did a benchmark, then compared it with a benchmark he made outside the freezer. The i9 was obviously throttling down at a reasonable room temperature. We suspect the i7 machines did the same. Long story short: some of the newer computers are being sold with new super-speedy hardware, but you can't use the extra computing power because the computer throttles down before you can use it.
(Score: 0) by Anonymous Coward on Friday October 18 2019, @07:34AM (1 child)
Consider using a primitive bash script that down throttles your CPU when the temp is too high. I can’t remember the name off the top of my head. The other option is to have a build system that uses Synth or Pourdier. If you opt for Synth, set the use flags in the make.conf and be explicit with the unset for the ‘choose one’ options. I suspect it’s a bug.
I opted for retired server hardware as my builder and haven’t looked back.
(Score: 0) by Anonymous Coward on Friday October 18 2019, @10:31AM
Subby is complaining that the BSD power management utility clocks down the CPU to prevent overheating under high load. Kitbashing a script to do the same thing only worse doesn't help.
I have a sneaking suspicion that there is nothing wrong with the laptop, but I would clean the fan and heat sinks just to be sure.
(Score: 1) by Sally_G on Friday October 18 2019, @07:35AM
https://www.dell.com/en-us/work/shop/accessories/apd/a1471445?cid=&st=Power__Cooling_Data_Center_Infrastructure&VEN1=,74079821731152,901q5c14135,c,,%7BProductid%7D&VEN2=bb,Power__Cooling_Data_Center_Infrastructure&lid=&dgc=st&dgseg=so&acd=12309152537461020&VEN3=811305150715096665 [dell.com]
Ugly URL, but I'll use it because I think Tripplite makes better quality than some other vendors. I'm surprised that anyone compiles on a laptop. Almost all desktops are better suited, and servers much more so, for the task. A cooling pad isn't going to cure your problem, but it will alleviate the problem, some.
(Score: 2) by GreatAuntAnesthesia on Friday October 18 2019, @08:59AM
I went to the doctor the other day,
I said 'it hurts when I do that'
he said 'well don't do it'
-- Tommy Cooper.
vid: https://www.youtube.com/watch?v=Kn4kH9c0JdA [youtube.com]
(Score: 2) by cockroach on Friday October 18 2019, @11:22AM
I used to run Gentoo Linux on my Thinkpad X201 and whenever I tried to upgrade Chromium, it would compile all night, then turn off due to heat once the CPU temperature was reaching about 100°C. Turns out, as others suggested, the main issue was dust. I removed the keyboard (which is very easy on these Laptops) and used one of those compressed air cans to dust off the cooler (spray on the fan from the inside, you will see a cloud of dust coming out of the air vent; also, turn off the machine when you do that, otherwise you will damage your fan).
That alone made the difference between overheating and merrily compiling Chromium all night long.
If this is not enough, changing the thermal paste on the CPU is also an option. The procedure a bit more complicated because you will have to completely remove the motherboard from the Laptop, but I had an X200 where new, proper paste made the difference between a loud, always-on fan and a nicely quiet machine.
(Score: 1) by zemm on Friday October 18 2019, @01:09PM
You can get laptop cooler pads with a fan (or multiple fans) that are designed to keep it cool. I haven't used one for many years but back in the day, the one I had did cool the laptop by a few degrees and it wasn't an expensive or top-of-the-line cooler.
(Score: 2) by Rich on Friday October 18 2019, @01:59PM
Not that any of my laptops (a merry mix of 2009-2012 MacBook Pros and ThinkPad Ts) has any of the indicated problems with my modest workloads, but...
I really have to get my cupboard-integrated fridge/freezer combo out and clean their radiators (or outright replace them). They've been sitting in their place since well before I got my place and must have accumulated a serious insulation layer of kitchen gunk by now. The wasted kWhs make not only Greta and the treehugger in me sad, but also my wallet.
(Score: 2) by microtodd on Friday October 18 2019, @02:27PM
I had a custom laptop with a powerful GPU for gaming. It overheats a lot.
Hard(ish) Solution: I opened it up and added a bunch of these (the small ones) to all the heat sink piping: https://smile.amazon.com/dp/B078KB7V5J [amazon.com]
(using thermal compound glue)
Easy solution: I bought something like this. It plugs into a usb port and a giant fan runs and keeps the laptop pretty cool.
https://smile.amazon.com/dp/B00NNMB3KS [amazon.com]
The cooler pad works well, its just noisy.
(Score: 2) by hwertz on Friday October 18 2019, @04:58PM
Not that it helps since you already have your computer... but I bought computers that are not space heaters.
Several vendors have decided both past and present "This CPU is so fast, it'll 'race to idle' so who cares if it's cooled properly." Well, yeah.
My current machine, I bought used and it turned out to have a 30W space heater in it; but it also is thick, heavy, and has plenty of cooling in it. My previous ones, an ARM Chromebook (typical power use 3W, 5W if you max the CPUs, maybe 10W if you max the CPUs and GPU -- but it's a NVidia GPU so "max out" means running CUDA code on it under Ubuntu, which basically never happened.) Before that I had a succession of Dells, but all 10-15W TDP. These stayed fairly cool. If you don't have 50W of processors in there, the machine does not overheat.
Other suggestion? Is there a *BSD version of thermald? This is an optional daemon in Linux distros, and works pretty great. If you have skin temp sensor, it tries to use that to keep the machine comfortable; otherwise you set a target temperature on some sensor it does have (i.e. cpu core temperature..) It seems to find real subtle cpu power saving modes, I ran it on a machine where the bottom would have stayed comfortable if I'd kept it like 6 (celsius, about 10 fahrenheit) cooler, about 20 degrees (celsius) cooler at the CPU cores. It found some subtle power save mode that somehow kept it that much cooler with 0 noticeable speed reduction. Set the core a degree or two cooler, then I started getting the 20-30% speed losses under load as expected as it resorted to thermal throttling. Alternately, write up a script to throttle or whatever only when the temp is high enough. You probably have stuff in /proc/ or /sys/ or somewhere to monitor temp from command line, and turn power save modes on and off to react to it.
(Score: 2) by srobert on Friday October 18 2019, @05:36PM
I have an old Acer laptop that runs FreeBSD. Overheating and shutdown occurs during large compilations. (I used to compile everything, the OS and applications from the ports tree). I used a sysctl command to limit the frequency of the CPU. It made those large compilations slower. But I always did those overnight anyway.
Since my wife started using it she prefers to just use the binary packages, so I haven't throttled it for a long while.
# sysctl -a | grep freq
should reveal some tunables and values that could be used to throttle it.
Do a search in the FreeBSD forums and you'll find the details.
(Score: 2, Informative) by paul_engr on Friday October 18 2019, @05:38PM (1 child)
I commend your dev efforts, but there's some low hanging fruit you should attack first.
1. Open it up and clean out the dust. Have an air compressor with blow gun or two cans of canned air ready. The heatsinks are thin fin and will be completely packed with hair and dust.
2. You can probably atet fan speed setpoints wrt/ temp in the BIOS
3. Make sure the laptop has clearance all around the vents so that the fans actually work and aren't just pushing air into an obstruction.
4. Get a cooler meant for laptop cooling. They're usually a box or lapdesk with fan ducts that hopefully blow into your fan inlets on the machine. Alternatively, you can sit ot up on a prop and blow a fan at the inlets.
(Score: 2, Informative) by paul_engr on Friday October 18 2019, @05:40PM
5. If the battery is bad and has high internal resistance, it will heat your machine up beyond the capacity of the cooling system. While it's running and plugged in, feel the pack. If you can remove it, pull it out and it should run fine on the power supply.
My old dells would heat my bedroom +5°F just idling because the pack was so bad and scorching hot.